I know this is made as a humorous video, but there is a lot more in the picture than the context gives.
PF is old. The framework hasn't had a rewrite since early 2015 (It's also way beyond the point of a possible rewrite and maintain all the functionality as all the server and client code together in PF is about almost 200k characters, which is about the equivalent of the first 3 harry potter books.) We even we hit lua's upvalue limit (which is 60 upvalues in the scope) and local limit (which is 200 locals in the scope) which prevented the proceeding of equipment. We eventually had to work around the upvalue limit and the frameworks quality went downhill.
Roblox has since made luau (Their own spin on lua) which raised the upvalue limit. But the framework is at the point if you add or change a line you are likely to break something or cause a obscure game breaking bug.
Currently in test place I'm restructuring a bunch of it to be in modulescripts, so later down the line we can better maintain and make additions and rewrite stuff. Currently it's hard to add content that won't require extra scripting but hopefully in the future this may be not a problem.

TL;DR: PF is getting hard to maintain. And it's code is about the size of 3 harry potter books, but in the future it may be possible to add content other than guns and attachments without everything breaking.

This was due to a oversight in the code, we forgot to make some values decay overtime. This has since been resolved in test place and is pending a merge to main game. The goal for it was to put you close to the battle but not too close. So you don't end up on the outskirts of mirage for example.

DerpCube It's easy for you to say. But you aren't doing the actual work. There is a difference in outcome between sitting on a lounge chair telling people what to do while not taking part in the actual work. Compared to somebody who is experienced and puts actual thought and consideration into their input.
The first results in people not considering your input and flat out ignoring you. The latter results in people actually considering your input and actually giving feedback on the circumstances.
So I'm giving you a suggestion, take a look at what you are saying and what I'm saying before I ignore you. I don't have to take your input. You are not entitled to my workflow.

VIP servers have every gun and attachment unlocked by default. - Courtesy of me.
(On another note, I'm going to work on adding more features to VIP servers in the near future. They have been pretty neglected.)

I can summarize what is wrong with PF in 2 words: Spaghetti code

Darkman_Bree Character limit is now 131068 characters.
(Don't crash the forums though, thanks.)

https://www.twitch.tv/stylis_studios
Talk about the stream on this thread.
(Also, First.)

Darkman_Bree I might eventually stream development, I probably won't be directly work on the game if I do though. Maybe smaller parts of the code in a separate instance to ensure nothing confidential in our code gets leaked. Cause when I work directly on the game I tend to hop all over the place looking at referencing implementations and editing existing implementations.

I win

fivefactor We are currently experimenting with a more proper system in test place currently. We don't place a player on a team internally until they spawn. This may cause issues with team spawning. But if this works well, we will find a workaround for that.

This does bring up a interesting issue, we currently have no way to tell if a user is AFK and discount them from the team balancing stuff.

If you want the factors used in the balancing it's essentially the following:
3 Hours worth of data maximum old records are thrown out
If the user has less than 15 minutes (1 Round) of data then we explicitly take their score as 0
And we generate a "score" of that 3 hours worth of rounds like with this example pseudo code

netkillsum = 0
playtimesum = 0

for round in 3hrs worth of rounds do
	netkillsum = netkillsum + round.kills - round.deaths
	playtimesum = playtimesum + round.playtimes

score = netkillssum/playtimesum

Then we sample that score for every player and put it in a hash table and sort it from greatest to least.
Then we process players 2 at a time, one will go on phantoms the other will go on ghosts. Each team will have a "absolute" score which would be a sum of all the team members scores.
Essentially it will guarantee each team will have at least one good player. Then after that it will try to keep the skill of each team somewhat similar.
Though that obviously doesn't account factors if a player is AFK or such.

PhantomSoldier You are forgetting about floating point error.
Edit: Besides roblox's walkspeed stuff isn't technically correct. Since the engine is step based you always have a margin of error due to the delta time.

Ikea Cat If we want to be technical here. A stud is sorta a dummy metric. It could be feet or meters or whatever you want. But internally roblox considers it as 1 stud = 1 foot
But in your own code it could be whatever technically.

dany47474747 Pinging him so he can look at it.

semaphorism

semaphorism

We have a few big problems:
#1 We don't have a sound guy anymore
#2 That is a lot of casings to make sounds for
#3 PF is a pile of junk and it would probably have a meltdown changing the code where that functionality resides.
Edit:
If we want to be specific here, we would have to make 76 different sounds to cover all the different calibers of casings in this game.

Outside of roblox? It made me worse.
Cause now I play more aggressively in almost every other FPS game.
Aggressive gameplay does not work in a game like siege...